latest and the best open source video generative ai models 2025

latest and the best open source video generative ai models 2025

With Analysis

With Analysis

With Analysis

text to video generative ai tools

Wan 2.2

WAN 2.2 is an advanced open source AI video generation model developed by Alibaba’s Tongyi Lab (WANX team) and released in 2025. It includes both Text‑to‑Video (T2V‑A14B) and Image‑to‑Video (I2V‑A14B) variants at 480p and 720p resolution, supporting 24 FPS, and optimized to run on consumer-grade GPUs like the RTX 4090 X (formerly Twitter) +11 Hugging Face +11 YouTube +11 . It’s open-source under Apache 2.0, integrates seamlessly with ComfyUI and Diffusers, and leverages a MoE architecture with prompt extension and multi-modal control features

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Wan 2.2

for the given prompts. Below are the discovered pros and cons for

Wan 2.2

.

Pros

Large community support: Integrated with ComfyUI and widely adopted across open source AI platforms.

Rich availability of workflows and tools: Easily accessible workflows and sample pipelines for WAN 2.2 across platforms like Hugging Face and GitHub.

Reliable text-to-video performance: Delivers impressive prompt fidelity for a fully open source AI model.

Strong adaptability: Performs well across a variety of prompt themes—from cinematic battle scenes to expressive character dialogue.

Cons

Heavy VRAM requirement for higher-resolution generation.

Output length limitations: Like many open source AI models, longer video generation is still under research and limitation.

Occasional frame inconsistency in complex or fast-paced motion scenes.

📝

Benchmark Score

⏲

Generation Speed

⚙️

ComfyUI Workflow

Wan Vace

Wan VACE is an advanced open source AI video generator developed by Alibaba’s Tongyi Lab in 2025. It supports text-to-video, image-to-video, and video editing tasks within one unified model. Available in 1.3B and 14B parameter sizes, Wan VACE offers high-quality AI video generation with multi-modal control like masks, poses, and flow. It's fully open-source under the Apache-2.0 license and used widely for research and creative applications in video generative AI.

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Wan Vace

for the given prompts. Below are the discovered pros and cons for

Wan Vace

.

Pros

All-in-one generation & editing: supports text-to-video, image-to-video, video-to-video, and masked editing in one unified model

Open-source and freely accessible with Apache‑2.0 license, hosted on GitHub, ModelScope, and Hugging Face

Multi-modal control via mask, pose, depth, flow, layouts — enhances prompt fidelity and creative direction

Performance-scaled sizes: 1.3 B model works on consumer GPUs (~8 GB VRAM), 14 B variant supports HD generation and advanced editing

Cons

Resource-intensive: full-scale 14 B version requires high compute (multi-GPU or premium hardware)

Complex setup: relies on command-line, Gradio, or ComfyUI — not plug-and-play for non-technical users

Documentation still growing: while core code and examples exist, full tutorial coverage is limited compared to commercial tools

Output consistency varies with prompt design and hardware; may need iterative refinement

📝

Benchmark Score

⏲

Generation Speed

Hunyuan AI

Hunyuan Video is a cutting-edge open source AI video generator developed by Tencent (Shenzhen-based tech giant) and released in December 2024. With 13 billion parameters, this video generative AI model uses advanced architectures like multimodal LLM text encoders, 3D VAE, and transformer-based diffusion to produce high-quality, cinematic videos up to 16 seconds in length at up to 1280×720p resolution. Fully open-source, it’s available on GitHub and Hugging Face, making it a benchmark in the ecosystem of open source AI video tools

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Hunyuan AI

for the given prompts. Below are the discovered pros and cons for

Hunyuan AI

.

Pros

Highest open‑source parameter count (~13B), resulting in strong visual quality, motion diversity, and text-video alignment

Supports text-to-video, image-to-video, and unified video workflows; extensible with LoRA and masks

Fully open-source (GitHub & Hugging Face), enabling community modification and research use

Flexible deployment: runs on local GPU via ComfyUI or multi‑GPU for faster inference; FP8 precision available

Cons

Significant compute requirements: 13B model needs high VRAM (12–16 GB+) and can take minutes per clip

Complex setup: command‑line and Gradio/ComfyUI integration—not user‑friendly for non‑technical users

Still improving documentation and tutorials; best practices can be scattershot

Generation time can be long (e.g., ~9 min for 3 sec at 720p, even on high‑end GPUs)

📝

Benchmark Score

⏲

Generation Speed

Ltxv 13b

Ltxv 13b is an open source AI video generator model developed by the LTX team and hosted on Hugging Face. It uses latent diffusion and transformer-based models to generate smooth, expressive videos from text and image prompts. Designed for creative experimentation, LTX‑V supports ComfyUI integration and allows seed-based variation for better prompt control. Its cinematic aesthetic and responsive prompt handling make it a go-to choice for developers and artists working with open source video generative AI models.

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Ltxv 13b

for the given prompts. Below are the discovered pros and cons for

Ltxv 13b

.

Pros

One of the fastest open source AI video generators available

Supports cinematic, stylized generation with consistent framing

Integrates easily with ComfyUI and Hugging Face workflows

Cons

Very limited motion in videos due to optimization that only regenerates moving pixels

Not ideal for prompts requiring complex movement or full-scene animation

📝

Benchmark Score

⏲

Generation Speed

Wan 2.2

WAN 2.2 is an advanced open source AI video generation model developed by Alibaba’s Tongyi Lab (WANX team) and released in 2025. It includes both Text‑to‑Video (T2V‑A14B) and Image‑to‑Video (I2V‑A14B) variants at 480p and 720p resolution, supporting 24 FPS, and optimized to run on consumer-grade GPUs like the RTX 4090 X (formerly Twitter) +11 Hugging Face +11 YouTube +11 . It’s open-source under Apache 2.0, integrates seamlessly with ComfyUI and Diffusers, and leverages a MoE architecture with prompt extension and multi-modal control features

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Wan 2.2

for the given prompts. Below are the discovered pros and cons for

Wan 2.2

.

Pros

Large community support: Integrated with ComfyUI and widely adopted across open source AI platforms.

Rich availability of workflows and tools: Easily accessible workflows and sample pipelines for WAN 2.2 across platforms like Hugging Face and GitHub.

Reliable text-to-video performance: Delivers impressive prompt fidelity for a fully open source AI model.

Strong adaptability: Performs well across a variety of prompt themes—from cinematic battle scenes to expressive character dialogue.

Cons

Heavy VRAM requirement for higher-resolution generation.

Output length limitations: Like many open source AI models, longer video generation is still under research and limitation.

Occasional frame inconsistency in complex or fast-paced motion scenes.

📝

Benchmark Score

⏲

Generation Speed

⚙️

ComfyUI Workflow

Wan Vace

Wan VACE is an advanced open source AI video generator developed by Alibaba’s Tongyi Lab in 2025. It supports text-to-video, image-to-video, and video editing tasks within one unified model. Available in 1.3B and 14B parameter sizes, Wan VACE offers high-quality AI video generation with multi-modal control like masks, poses, and flow. It's fully open-source under the Apache-2.0 license and used widely for research and creative applications in video generative AI.

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Wan Vace

for the given prompts. Below are the discovered pros and cons for

Wan Vace

.

Pros

All-in-one generation & editing: supports text-to-video, image-to-video, video-to-video, and masked editing in one unified model

Open-source and freely accessible with Apache‑2.0 license, hosted on GitHub, ModelScope, and Hugging Face

Multi-modal control via mask, pose, depth, flow, layouts — enhances prompt fidelity and creative direction

Performance-scaled sizes: 1.3 B model works on consumer GPUs (~8 GB VRAM), 14 B variant supports HD generation and advanced editing

Cons

Resource-intensive: full-scale 14 B version requires high compute (multi-GPU or premium hardware)

Complex setup: relies on command-line, Gradio, or ComfyUI — not plug-and-play for non-technical users

Documentation still growing: while core code and examples exist, full tutorial coverage is limited compared to commercial tools

Output consistency varies with prompt design and hardware; may need iterative refinement

📝

Benchmark Score

⏲

Generation Speed

Hunyuan AI

Hunyuan Video is a cutting-edge open source AI video generator developed by Tencent (Shenzhen-based tech giant) and released in December 2024. With 13 billion parameters, this video generative AI model uses advanced architectures like multimodal LLM text encoders, 3D VAE, and transformer-based diffusion to produce high-quality, cinematic videos up to 16 seconds in length at up to 1280×720p resolution. Fully open-source, it’s available on GitHub and Hugging Face, making it a benchmark in the ecosystem of open source AI video tools

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Hunyuan AI

for the given prompts. Below are the discovered pros and cons for

Hunyuan AI

.

Pros

Highest open‑source parameter count (~13B), resulting in strong visual quality, motion diversity, and text-video alignment

Supports text-to-video, image-to-video, and unified video workflows; extensible with LoRA and masks

Fully open-source (GitHub & Hugging Face), enabling community modification and research use

Flexible deployment: runs on local GPU via ComfyUI or multi‑GPU for faster inference; FP8 precision available

Cons

Significant compute requirements: 13B model needs high VRAM (12–16 GB+) and can take minutes per clip

Complex setup: command‑line and Gradio/ComfyUI integration—not user‑friendly for non‑technical users

Still improving documentation and tutorials; best practices can be scattershot

Generation time can be long (e.g., ~9 min for 3 sec at 720p, even on high‑end GPUs)

📝

Benchmark Score

⏲

Generation Speed

Ltxv 13b

Ltxv 13b is an open source AI video generator model developed by the LTX team and hosted on Hugging Face. It uses latent diffusion and transformer-based models to generate smooth, expressive videos from text and image prompts. Designed for creative experimentation, LTX‑V supports ComfyUI integration and allows seed-based variation for better prompt control. Its cinematic aesthetic and responsive prompt handling make it a go-to choice for developers and artists working with open source video generative AI models.

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

After using and testing

Ltxv 13b

for the given prompts. Below are the discovered pros and cons for

Ltxv 13b

.

Pros

One of the fastest open source AI video generators available

Supports cinematic, stylized generation with consistent framing

Integrates easily with ComfyUI and Hugging Face workflows

Cons

Very limited motion in videos due to optimization that only regenerates moving pixels

Not ideal for prompts requiring complex movement or full-scene animation

📝

Benchmark Score

⏲

Generation Speed

Wan 2.2

WAN 2.2 is an advanced open source AI video generation model developed by Alibaba’s Tongyi Lab (WANX team) and released in 2025. It includes both Text‑to‑Video (T2V‑A14B) and Image‑to‑Video (I2V‑A14B) variants at 480p and 720p resolution, supporting 24 FPS, and optimized to run on consumer-grade GPUs like the RTX 4090 X (formerly Twitter) +11 Hugging Face +11 YouTube +11 . It’s open-source under Apache 2.0, integrates seamlessly with ComfyUI and Diffusers, and leverages a MoE architecture with prompt extension and multi-modal control features

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

Pros

Large community support: Integrated with ComfyUI and widely adopted across open source AI platforms.

Rich availability of workflows and tools: Easily accessible workflows and sample pipelines for WAN 2.2 across platforms like Hugging Face and GitHub.

Reliable text-to-video performance: Delivers impressive prompt fidelity for a fully open source AI model.

Strong adaptability: Performs well across a variety of prompt themes—from cinematic battle scenes to expressive character dialogue.

Cons

Heavy VRAM requirement for higher-resolution generation.

Output length limitations: Like many open source AI models, longer video generation is still under research and limitation.

Occasional frame inconsistency in complex or fast-paced motion scenes.

📝

Benchmark Score

⏲

Generation Speed

⚙️

ComfyUI Workflow

Wan Vace

Wan VACE is an advanced open source AI video generator developed by Alibaba’s Tongyi Lab in 2025. It supports text-to-video, image-to-video, and video editing tasks within one unified model. Available in 1.3B and 14B parameter sizes, Wan VACE offers high-quality AI video generation with multi-modal control like masks, poses, and flow. It's fully open-source under the Apache-2.0 license and used widely for research and creative applications in video generative AI.

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

Pros

All-in-one generation & editing: supports text-to-video, image-to-video, video-to-video, and masked editing in one unified model

Open-source and freely accessible with Apache‑2.0 license, hosted on GitHub, ModelScope, and Hugging Face

Multi-modal control via mask, pose, depth, flow, layouts — enhances prompt fidelity and creative direction

Performance-scaled sizes: 1.3 B model works on consumer GPUs (~8 GB VRAM), 14 B variant supports HD generation and advanced editing

Cons

Resource-intensive: full-scale 14 B version requires high compute (multi-GPU or premium hardware)

Complex setup: relies on command-line, Gradio, or ComfyUI — not plug-and-play for non-technical users

Documentation still growing: while core code and examples exist, full tutorial coverage is limited compared to commercial tools

Output consistency varies with prompt design and hardware; may need iterative refinement

📝

Benchmark Score

⏲

Generation Speed

Hunyuan AI

Hunyuan Video is a cutting-edge open source AI video generator developed by Tencent (Shenzhen-based tech giant) and released in December 2024. With 13 billion parameters, this video generative AI model uses advanced architectures like multimodal LLM text encoders, 3D VAE, and transformer-based diffusion to produce high-quality, cinematic videos up to 16 seconds in length at up to 1280×720p resolution. Fully open-source, it’s available on GitHub and Hugging Face, making it a benchmark in the ecosystem of open source AI video tools

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

Pros

Highest open‑source parameter count (~13B), resulting in strong visual quality, motion diversity, and text-video alignment

Supports text-to-video, image-to-video, and unified video workflows; extensible with LoRA and masks

Fully open-source (GitHub & Hugging Face), enabling community modification and research use

Flexible deployment: runs on local GPU via ComfyUI or multi‑GPU for faster inference; FP8 precision available

Cons

Significant compute requirements: 13B model needs high VRAM (12–16 GB+) and can take minutes per clip

Complex setup: command‑line and Gradio/ComfyUI integration—not user‑friendly for non‑technical users

Still improving documentation and tutorials; best practices can be scattershot

Generation time can be long (e.g., ~9 min for 3 sec at 720p, even on high‑end GPUs)

📝

Benchmark Score

⏲

Generation Speed

Ltxv 13b

Ltxv 13b is an open source AI video generator model developed by the LTX team and hosted on Hugging Face. It uses latent diffusion and transformer-based models to generate smooth, expressive videos from text and image prompts. Designed for creative experimentation, LTX‑V supports ComfyUI integration and allows seed-based variation for better prompt control. Its cinematic aesthetic and responsive prompt handling make it a go-to choice for developers and artists working with open source video generative AI models.

PROMPT 1

"A hyperrealistic video of an interview in a professional TV studio. A tall Indian basketball player, nearly 3 meters tall, sits on a modern chair, dressed in a tailored suit with subtle basketball-themed details. He is visibly towering, with expressive brown eyes and an athletic build. Next to him sits a professional female journalist with dark hair and stylish glasses, holding a microphone and a notebook. The journalist asks questions with an engaging, confident demeanor. Both appear under soft studio lighting, with a sleek background featuring a subtle basketball motif. The atmosphere is serious yet friendly, showing close-up shots of their faces, hands, and body language. The scene feels cinematic, realistic, and high-definition."

Show Observations

Pros

One of the fastest open source AI video generators available

Supports cinematic, stylized generation with consistent framing

Integrates easily with ComfyUI and Hugging Face workflows

Cons

Very limited motion in videos due to optimization that only regenerates moving pixels

Not ideal for prompts requiring complex movement or full-scene animation

📝

Benchmark Score

⏲

Generation Speed

benchmark scores of open source Video generation ai models

To check the methodology behind testing of AI models, click here

AI Model
Prompt Adherence
Realism
Frame Stability
Motion Quality
Identity Consistency
Scene Continuity
Physics & Interactions
Camera & Lighting
WAN 2.1 VACE
★★★½☆
★★★½☆
★★★★☆
★★★½☆
★★★½☆
★★★½☆
★★★☆☆
★★★½☆
Hunyuan Video
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★½☆
★★★★☆
LTX-V 1.3B
★★★☆☆
★★½☆☆
★★★☆☆
★★½☆☆
★★½☆☆
★★½☆☆
★★☆☆☆
★★★☆☆
WAN 2.2
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★½☆
★★★★☆
AI Model
Prompt Adherence
Realism
Frame Stability
Motion Quality
Identity Consistency
Scene Continuity
Physics & Interactions
Camera & Lighting
WAN 2.1 VACE
★★★½☆
★★★½☆
★★★★☆
★★★½☆
★★★½☆
★★★½☆
★★★☆☆
★★★½☆
Hunyuan Video
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★½☆
★★★★☆
LTX-V 1.3B
★★★☆☆
★★½☆☆
★★★☆☆
★★½☆☆
★★½☆☆
★★½☆☆
★★☆☆☆
★★★☆☆
WAN 2.2
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★½☆
★★★★☆
AI Model
Prompt Adherence
Realism
Frame Stability
Motion Quality
Identity Consistency
Scene Continuity
Physics & Interactions
Camera & Lighting
WAN 2.1 VACE
★★★½☆
★★★½☆
★★★★☆
★★★½☆
★★★½☆
★★★½☆
★★★☆☆
★★★½☆
Hunyuan Video
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★½☆
★★★★☆
LTX-V 1.3B
★★★☆☆
★★½☆☆
★★★☆☆
★★½☆☆
★★½☆☆
★★½☆☆
★★☆☆☆
★★★☆☆
WAN 2.2
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★★☆
★★★½☆
★★★★☆

Time taken by open source ai model to generate a 720p 24 fps 5 seconds video on a 16 gb vram RTX 5070 TI graphic card

AI Model
Time Taken
Wan Vace
25-30 mins
Hunyuan AI
30-35 mins
Ltxv 13b
45 secs
Wan 2.2
20-25 mins
AI Model
Time Taken
Wan Vace
25-30 mins
Hunyuan AI
30-35 mins
Ltxv 13b
45 secs
Wan 2.2
20-25 mins
AI Model
Time Taken
Wan Vace
25-30 mins
Hunyuan AI
30-35 mins
Ltxv 13b
45 secs
Wan 2.2
20-25 mins